Claude Cowork: Automate 99% of Your Life (New Meta)

Jack Roberts15m 6s

Jack Roberts introduces Claude Co-work's new Projects feature, which solves the 'disposable thread' problem by giving Claude persistent memory and context for specific business areas. He demonstrates how to build a personal operating system using projects with connectors, skills, and organized workspaces.

Summary

Jack Roberts, who built and sold a tech startup with 60,000 customers, explains Claude Co-work's revolutionary Projects feature that transforms how users interact with AI. The main problem this solves is the 'disposable thread' issue where AI loses context after conversations end, forcing users to constantly repeat themselves. Projects creates persistent intelligence that gets smarter over time by maintaining four key components: instructions (brand voice, rules, tone), context (knowledge base, folders, URLs), memory (automatic scope per project), and scheduled tasks. Roberts demonstrates the setup process, showing how to create a YouTube project with specific instructions, file uploads, and connectors to various tools like Notion, Google Calendar, and Canva. He emphasizes connecting your world through integrations and building skills for repetitive tasks like invoice generation. The video covers decision-making criteria for creating projects: tasks repeated weekly, things constantly re-explained to AI, and workflows with multiple moving parts. Roberts also clarifies the differences between folders (simple file access without memory), projects (full featured with memory and instructions), and skills (portable functions that work everywhere). He concludes by explaining the difference between Claude Co-work and Claude Code, noting that Co-work has a better UI and sharing capabilities but may use tokens faster.

Key Insights

  • Roberts argues that AI conversations are 'disposable threads' that lose context due to finite context windows, causing users to constantly repeat themselves and waste time
  • Roberts claims that Claude Projects solves the 'continuity problem' by giving Claude a persistent brain that gets smarter with more use, transforming it from a tool into a workspace that knows you and your business
  • Roberts states that Projects have four essential components: instructions for brand voice and rules, context as knowledge base, memory with automatic scope per project, and scheduled tasks for continuous operations
  • Roberts recommends using the prompt 'Based on our entire conversation, if you had to create buckets for our topics, for projects within co-work, what would they be?' to let Claude suggest project categories based on conversation history
  • Roberts explains that Claude Code can do everything Claude Co-work can do, but Co-work has better UI for documents and sharing capabilities with teams, though it may burn tokens faster

Topics

Claude Co-work ProjectsAI automationpersistent memorybusiness workflowsconnectors and integrationspersonal operating systems

Full transcript available for MurmurCast members

Sign Up to Access

Get AI summaries like this delivered to your inbox daily

Get AI summaries delivered to your inbox

MurmurCast summarizes your YouTube channels, podcasts, and newsletters into one daily email digest.